About a year ago (and perhaps once a month ever since) Markus Gobel asked: “Why is it still impossible to do a hosted-Skype-to-SIP?”
He suggested that a service provider should install servers to act as a bridge to Skype where users give the service provider their Skype logins. The reason this is not common is because of cost. To bridge to open-standards based VoIP, particularly SIP, is easy. The protocols are known, and there are many choices for implementations to deploy. Skype, on the other hand, is proprietary and there is only one source for an implementation, Skype. It is far more resource intensive and complex to setup “hosted Skype”.
A modern server can support thousands (or hundreds of thousands, perhaps millions) of SIP accounts, and it is easy to manage such configurations using off-the-shelf solutions. Contrast that to Skype, where running “hosted Skype” requires all kinds of hacks. One has to run instances of the normal Skype client with some kind of hacks to a PBX or SIP or other open system. How many hacked Skype instances can you run on one server? Are you managing these in a pile of virtual machines? Is each VM running Windows? If so, that means expensive Microsoft licenses per virtual machine. This is ugly and complex, which translates to expensive to setup and operate.
Mobivox and Fring must be deploying such hacks – and it must be expensive and complex operationally for them to do so.
And that’s why “hosted Skype” is not common. If Skype had a published interface (we call them protocols) to interface at the network level, we would see third-party code that could be used on the platform(s) of the operator’s choice. The Skype protocol might still turn out to be more expensive to operate per user (have more overhead) for the operator than standards-based options, but it would be far less expensive than the hacks required today to deploy “hosted Skype”.
Even if Skype offered “black box” (closed) modules with APIs for a server environment, it would be better than the situation we have today. Instead, today we have to work with Skype clients intended to interact with humans, with limited APIs, that have to be “faked” into data center mode – we have no “dark data center” human-free Skype modules to work with.
So don’t blame the operators and service providers – blame the cause of the situation SKYPE! Skype needs to provide “data center” solutions if you expect more operators and service providers to deploy “hosted Skype” services.
another aspect that almost no SIP company admits too is that they are in many cases are not proxing the calls through there servers at all. at least not the voice stream which is going directly point to point over the internet. that is why companies like mobivox, fring, net2max, and some of the flash based VOIP companies that cliam to have skype operability in the plans are the ones that do offer skype interop. since 100% of the calls they handle go through their servers anyway.
Tom, you say that like it’s a bad thing, as if a company would make a secret of it. What is wrong with having the voice stream take the more efficient path, going directly point to point over the internet? And why would a company not want to "admit" that? It seems like something to brag about to me.
sorry. i guess you misunderstood me. what i have noticed though is that smaller VOIP startups will beign there services with proxing audio; later when they move away from that customers behind NAT will start to have problems with one way audio, etc.
i am not going to point at any specific companies; but i can tell you from experience they are not always so willing to admit that the audio and/or connection problems customers suddenly start to havce are a result of the elimination of there servers.
there are significant advatages to both direct peer2peer connections and proxied connections. it would be nice if as consumers we could ask our services providers how calls are proccessed. many VOIP companies do not answer that question when asked.
David, Nice piece… it turns out that you and I were blogging along similar lines. I wrote this piece Jan 15th:
http://www.disruptivetelephony.com/2008/01/skype-says-no-t.html
I was pleased to see Jonathan Christensen from Skype talk about the benefits of SIP in his recent interview for EComm 2008:
http://www.disruptivetelephony.com/2008/02/the-ecomm-2008.html
In that interview, he talks about how "walled gardens won’t survive" which is rather interesting given Skype’s continuing lack of interoperability.
It’s very clear that pieces of the *infrastructure* are there to do SIP-to-Skype interop. As Christensen says in the interview, Skype is a huge user of SIP connections. They just need to figure out how to make that connectivity available to users.
Look at SippySkype at http://www.mhspot.com – free open source skype to sip bridge.
Thanks Greg. SippySkype looks cool. However, it’s not a "hosted service" – it runs along side a Skype client that you have to run on some machine, which is exactly what Markus Gobel (the catalyst for my post) is trying to avoid, as am I. He says here http://www.toyz.org/mrblog/archives/00000369.html#comments that he wants hosted Skype to SIP so that he "would never have to switch on Skype again." … "on Skype I am nearly never reachable because I don’t like communications that force me to switch my computer on."
SippySkype is an example of the kinds of what I speak about, of coercing a GUI app (Skype) to act like a gateway/server – a cool hack, but people want a service for it, not to dedicate a computer runnigng Skype to this task.
How about this: http://www.netgear.com/Products/CommunicationsVoIP/Skype/SPH200D.aspx
Claims no computer needed.
The Sip to Skype hardware gateways out there are expensive. You might as weel leave a computer on all the time.
That’s kind of the point, Greg. People like Markus don’t WANT to operate their own hardware gateway OR to leave a computer on. And they don’t want another phone (like the Netgear).
Your point that hardware Skype gateways are expensive confirms what I’m saying – such gateways are expensive because they all have to be hacks, "tricking" Skype into doing something it doesn’t really want to do (that’s what makes them expensive) and that’s why we don’t see hosted services employing such gateways for Skype, which is what I said in the first place.